Download 1 - Institut Jean Nicod

The limits of expressibility
François Recanati
CNRS, Paris
([email protected])
To appear in:
B. Smith (ed.), John Searle
Cambridge University Press
I. The determination view
A basic tenet of contemporary semantics is that the meaning of a sentence
determines its truth conditions. That determination of truth conditions by
linguistic meaning can be more or less direct. In non-indexical cases the
meaning of the sentence directly determines (and can even be equated
with) its truth conditions. The sentence 'Snow is white' means that, and is
true if and only if, snow is white. To give the meaning of such a sentence
is to give its truth conditions. When the sentence is indexical, the situation
is more complex. The sentence has truth conditions only 'with respect to
context'. The sentence 'I am English' uttered by John is true if and only if
John is English; uttered by Paul, it is true if and only if Paul is English, etc.
Still, the truth conditions, in each context, are determined by the meaning
of the sentence (with respect to the context). The meaning of 'I am English'
determines that, if that sentence is uttered by a, then it is true iff a is
English.
The thesis that meaning determines truth conditions can be dubbed
the 'Determination View'. It goes largely unquestioned in contemporary
analytic philosophy (as it did in early analytic philosophy). Fifty years ago
the situation was different. So-called 'ordinary language philosophers'
rejected the Determination View.1 But ordinary language philosophy has
suffered a spectacular loss of influence over the last thirty years and is
nowadays no more than an object of scorn and caricature. The interest
1
Amongst 'ordinary language philosophers' I include Wittgenstein and Waismann as
well as Oxford philosophers (Austin, Strawson, etc.).
aroused by Wittgenstein and his work has not, paradoxically, ceased to
grow even though the current of ideas from which he is inseparable has
undergone the aforesaid decline. But Wittgenstein's more or less
intentional obscurity, even if contributing to his impact and popularity,
limits the effective dissemination of his ideas.
Among contemporary theorists, only a few resist the Determination
View. John Searle is one of them. He holds that linguistic meaning
radically under-determines truth conditions. According to Searle, even
after the reference or semantic value of all the indexical expressions
contained in a sentence, including tenses, has been contextually fixed, we
still cannot specify a state of affairs s such that the sentence (with respect
to those contextual assignments) is true if and only if s obtains. For every
candidate, i.e. for every such state of affairs, Searle shows that we can
imagine a context with respect to which the sentence would not, or not
necessarily, be considered as true, even though the relevant state of affairs
obtained. That is so because in specifying the state of affairs in question
we take many things for granted. If we get rid of those tacit assumptions
(by imagining a weird context in which they do not hold) the state of
affairs we have specified no longer corresponds to the intuitive truth
conditions of the utterance.
II. Background assumptions
There are many things that we take for granted both in speaking and in
interpreting the utterances of others. Among those things we take for
granted, some are articulated in the sentence itself: they are the
'presuppositions' of the sentence. Thus if I say that John has stopped
smoking, I presuppose that John smoked before, in virtue of the
appropriateness conditions of the verb 'to stop'. But there are also things
we take for granted which are in no way articulated in the sentence itself.
Searle calls them 'background assumptions'. For example,
Suppose I go into the restaurant and order a meal. Suppose I say,
speaking literally, 'Bring me a steak with fried potatoes.' (...) I take it
for granted that they will not deliver the meal to my house, or to my
place of work. I take it for granted that the steak will not be encased
in concrete, or petrified. It will not be stuffed into my pockets or
spread over my head. But none of these assumptions was made
explicit in the literal utterance. (Searle 1992: 180)
Though unarticulated, those assumptions contribute to determining the
intuitive conditions of satisfaction (obedience conditions, truth conditions,
etc.) of the utterance. The order 'Bring me a steak with fried potatoes' does
not count as satisfied if the steak is delivered, encased in concrete, to the
customer's house. It is mutually manifest to both the hearer and the speaker
that the speaker intends the ordered meal to be placed in front of him on
the restaurant table he is sitting at, etc. Though not explicitly said, that is
clearly part of what is meant. Yet one does not want to say that that aspect
of utterance meaning is conveyed indirectly or nonliterally (as when one
says something and means something else). The utterance 'Bring me a
steak with fried potatoes' is fully literal. It is a property of literal and
serious utterances that their conditions of satisfaction systematically
depend upon unstated background assumptions.
Another example given by Searle involves the word 'cut' in (literal
utterances of) sentences such as 'Bill cut the grass' and 'Sally cut the cake'.
The word 'cut' is not ambiguous, Searle says, yet it makes quite different
contributions to the truth conditions of the utterance in the two cases. That
is because background assumptions play a role in fixing satisfactionconditions, and different background assumptions underlie the use of 'cut'
in connection with grass and cakes respectively. We assume that grass is
cut in a certain way, and cakes in another way. The assumed way of
cutting finds its way into the utterance's truth conditions:
Though the occurrence of the word "cut" is literal in [both]
utterances..., and though the word is not ambiguous, it determines
different sets of truth conditions for the different sentences. The sort
of thing that constitutes cutting the grass is quite different from, e.g.,
the sort of thing that constitutes cutting a cake. One way to see this is
to imagine what constitutes obeying the order to cut something. If
someone tells me to cut the grass and I rush out and stab it with a
knife, or if I am ordered to cut the cake and I run over it with a
lawnmower, in each case I will have failed to obey the order. That is
not what the speaker meant by his literal and serious utterance of the
sentence. (Searle 1980: 222-223)
Examples can be multiplied at will. Searle convincingly shows that
background assumptions such as these have the following properties: (i)
for every utterance, there is an indefinite number of them; (ii) if we
manipulate them by imagining weird situations in which they do not hold
(e.g. a situation in which steaks are standardly encased in concrete, or a
situation in which grass is sliced into strips and sold to customers who
want a lawn in a hurry), the intuitive truth conditions of the utterance are
affected; (iii) we cannot make them explicit in the sentence itself without
bringing in further background assumptions involved in the interpretation
of the extra descriptive material. These properties entail that the
Determination View must be given up. Truth-conditions depend not only
upon the meaning of the sentence and the contextual parameters relevant
to the interpretation of indexicals, but also upon what Searle calls 'the
Background'. Change the background, you change (or possibly destroy)
the truth conditional content of the utterance.
III. Contextualism
As I mentioned above, ordinary language philosophers, from Wittgenstein
to Strawson, also rejected the Determination View. Their emphasis was on
speech rather than language. As Searle writes,
[Early analytic philosophers] treat the elements of language — words,
sentences, propositions — as things that represent or things that are
true or false, etc., apart from any actions or intentions of speakers and
hearers. The elements of the language, not the actions and intentions
of the speakers are what count. In the late thirties and especially after
the Second world War these assumptions came to be vigourously
challenged, especially by Wittgenstein. (Searle 1971: 6)
According to the alternative view put forward by Wittgenstein, Austin, and
their followers, it is not natural language sentences, not even sentences
'with respect to context', which have truth conditions, but full-blooded
speech acts — meaningful actions performed by rational agents. That view
I call 'contextualism'.
Two purported refutations of contextualism were offered in the
sixties, by Grice and Geach respectively. One of the reasons why
contextualism has lost grounds is that those refutations have been
generally considered as successful. Another reason for the demise of
contextualism is the striking success of the intellectual entreprise known
as formal semantics. Formal semantics is based on the Determination
View, and its success seems to give the lie to contextualism.
In this chapter I will not deal with Grice's and Geach's arguments
against contextualism;2 nor will I consider whether or not it is possible to
reconcile contextualism with the project of giving a systematic semantics
for natural language. I will only be concerned with Searle's critique of the
Determination View, and its relation (both historical and theoretical) to
contextualism.
Even though he was trained in Oxford under Austin and Strawson,
Searle himself was not a contextualist when he wrote Speech Acts. To be
sure, he held that "the unit of communication is not, as has generally been
supposed, the symbol, word or sentence, or even the token of the symbol,
word or sentence, but rather the production or issuance of the symbol or
word or sentence in the performance of the speech act" (Searle 1965: 221222; 1969: 16). At the same time, however, he issued warnings such as the
following:
A commonplace of recent philosophizing about language has been the
distinction between sentences and the speech acts performed in the
utterances of those sentences. Valuable as this distinction is, there has
also been a tendency to overemphasize it. (Searle 1968: 153)
It is possible to distinguish at least two strands in contemporary work
in the philosophy of language — one which concentrates on the uses
of expressions in speech situations and one which concentrates on the
meaning of sentences. Practitioners of these two approaches
sometimes talk as if they were inconsistent, and at least some
encouragement is given to the view that they are inconsistent by the
fact that historically they have been associated with inconsistent
views about meaning. (...) But although historically there have been
sharp disagreements between practitioners of these two approaches, it
is important to realize that the two approaches... are complementary
and not competing. (Searle 1969: 18)
The main reason why Searle kept his distances from the contextualism of
his teachers was his acceptance of a basic principle he put forward under
the name of 'Principle of Expressibility'.
2
For a critique of Grice's argument against contextualism, see Recanati 1994.
IV. The Principle of Expressibility
In general, the content of a speech act — what the speaker communicates
and the hearer understands — cannot be equated with the content of the
sentence uttered in performing that speech act. That is due to many factors.
(i) The uttered sentence is often elliptical, indeterminate or ambiguous
even though what the speaker communicates by uttering the sentence in
context is perfectly determinate and univocal. (ii) The referring
expressions used by the speaker do not, in general, uniquely determine
what the speaker is referring to: appeal to speaker's intentions is necessary
to fix the reference of, say, demonstratives pronouns etc. (iii) Beside what
she says, there are many things that the speaker conveys implicitly or
nonliterally by her utterance, as in indirect speech acts, irony and
metaphor. The three factors result in a gap between literal sentence
meaning and speaker's utterance meaning. But that gap can always be
closed: that is the gist of the Principle of Expressibility, according to
which whatever can be meant can be said. In principle if not in fact, it is
always possible to utter a fully explicit sentence, that is, a sentence whose
linguistic meaning exactly corresponds to, and uniquely determines, the
force-and-content of the speech act one is performing. It follows that "a
study of the meaning of sentences is not in principle distinct from a study
of speech acts. Properly construed, they are the same study" (Searle 1969:
18).
Interpreted at face value, the Principle of Expressibility is
incompatible with contextualism. According to contextualism, the sort of
content which utterances have (in virtue of the speech acts they serve to
perform) can never be fully encoded into a sentence; hence it will never be
the case that the sentence itself expresses that content in virtue solely of
the conventions of the language. Sentences, by themselves, do not have
determinate contents. What gives them the determinate contents they have
(in context) is the fact that they are used in performing meaningful actions.
In brief, contextualism says that the gap between sentence meaning and
speaker's meaning can never be closed; while the Principle of
Expressibility says it can always be closed. A consequence of the Principle
of Expressibility, Searle says, is that "cases where the speaker does not say
exactly what he means — the principal kinds of cases of which are
nonliteralness, vagueness, ambiguity, and incompleteness — are not
theoretically essential to linguistic communication" (Searle 1969: 20).
According to contextualism, however, the under-determination of
communicated content by linguistic meaning is an essential feature of
linguistic communication.
Just as it is incompatible with contextualism, the Principle of
Expressibility seems incompatible with Searle's findings about
background-dependence. For Searle, as much as for Austin or
Wittgenstein, linguistic meaning essentially under-determines
communicated content. As we have seen, it is impossible to make the
background assumptions against which an utterance is interpreted explicit,
first because there is an indefinite number of such assumptions, and
second because one cannot make them explicit without bringing in further
background assumptions against which the extra descriptive material is
interpreted. It follows that the content communicated by an utterance
cannot be fully encoded into the sentence.
Yet Searle has explicitly denied that background phenomena
threaten the Principle of Expressibility. In 'Literal Meaning', he writes:
"There is nothing in the thesis of the relativity of literal meaning which is
inconsistent with the Principle of Expressibility, the principle that
whatever can be meant can be said" (Searle 1979: 134). How are we to
make sense of that denial?
V. Expressibility and Effability
Searle's Principle is incompatible with contextualism and the thesis of
background-dependence when interpreted at face-value. But Searle's
formulations are vague, and it is possible, if somewhat strained, to
distinguish several possible interpretations for the Principle.
On the strongest, and most natural, interpretation — that which I
took for granted so far — Searle's Principle of Expressibility is equivalent
to Katz's Principle of Effability (Katz 1972: 18-24). Katz defines the
(grammatical) meaning of a sentence as the meaning it has in the 'null
context'; and he says that what the speaker means by uttering a sentence S
in a context C (the 'utterance meaning' of S) can always be made explicit
as the 'grammatical meaning' of an alternative sentence S' that the speaker
might have uttered (Principle of Effability). Katz says his Principle "was
propounded in somewhat different form by... Searle"3 in passages such as
the following:
3
Katz 1978: 209.
If you ask me "Are you going to the movies?" I may respond by
saying "Yes" but, as is clear from the context, what I mean is "Yes, I
am going to the movies", not "Yes, it is a fine day" or "Yes, we have
no bananas". Similarly, I might say "I'll come" and mean it as a
promise to come, i.e., mean it as I would mean "I promise that I will
come", if I were uttering that sentence and meaning literally what I
say. In such cases, even though I do not say exactly what I mean, it is
always possible for me to do so — if there is any possibility that the
hearer might not understand me, I may do so. (Searle 1969: 19)
One possible difference between Searle's and Katz's respective
principle lies in Katz's appeal to the notion of 'null context' in
characterizing grammatical meaning (a notion which Searle later
criticized). Where Katz invokes the distinction between grammatical
meaning thus characterized and utterance meaning, Searle appeals to a
vaguer distinction between 'sentence meaning' and 'intended speaker
meaning'. Not only is that distinction vague, it is ambiguous in Searle's
writings (Recanati 1987: 255-6). In 'Austin on locutionary and
Illocutionary Acts', Searle says that sense and reference are two "of the
aspects... in which intended speaker-meaning may go beyond literal
sentence-meaning" (Searle 1968: 149). Here Searle presumably identifies
literal sentence meaning with the linguistic, 'determinable' meaning of the
sentence-type, and intended speaker meaning with what the speaker says
in uttering this sentence (cf. Forguson, 1973: 179). That is more or less the
same distinction as Katz's distinction between grammatical meaning and
utterance meaning, or Austin's distinction between 'phatic' meaning and
'rhetic' meaning. But in 'Indirect Speech Acts' (and again in 'Metaphor'),
what Searle calls 'sentence meaning' is what the speaker literally says (by
uttering the sentence in context), and what he calls 'speaker's utterance
meaning' is what the speaker actually conveys or communicates (which
may, and typically does, go beyond or otherwise diverge from what is
said). Table 1 (from Recanati 1980: 206) summarizes Searle's ambiguous
use of the sentence meaning/speaker's meaning distinction.
Linguistic
meaning of the
sentence type
Searle 1968
Searle 1975
sentence
meaning1
what is literally
said by uttering
the sentence in
context
speaker's
meaning1
sentence
meaning2
what is thereby
communicated
speaker's
meaning2
Table 1
Given that ambiguity, it is tempting to substantiate Searle's claim
that the Principle of Expressibility is consistent with his later findings by
interpreting the Principle as follows:
The gap between sentence meaning2 and speaker's meaning2 can
always be closed. In other words, what is implied or indirectly
conveyed can always be said literally or directly conveyed. But what
the sentence says — its literal content (sentence meaning2) — can
still be treated as context-relative and background-dependent.
On that interpretation of the Principle of Expressibility, it is no longer
entailed that the content of every speech act can be fully encoded into the
linguistic meaning of a sentence type.
That interpretation of the Principle of Expressibility is clearly not
what Searle intended when he wrote the relevant passages in his early
works, however. It is not only nonliteralness, but all cases of divergence
between sentence meaning and speaker utterance meaning, including (inter
alia) 'vagueness, ambiguity, and incompleteness', which he says are not
theoretically essential to linguistic communication, in virtue of the
Principle of Expressibility. There does not seem to be any significant
difference between Searle's Principle of Expressibility and Katz's Principle
of Effability in that respect. In particular, there is no reason to think that
Searle would have disapproved of anything in the following passage in
which Katz talks about the divergence between grammatical meaning and
utterance meaning:
Given that the utterance meaning of a sentence S can be expressed as
the grammatical meaning of another sentence S', why isn't our
performance mechanism designed to use S' in the first place? What
purpose is served by having it produce S and depend on information
about the context to supply the hearers with part of the utterance
meaning of S? One function performed by such a mechanism is to
increase our repertoire of verbal behavior by permitting us to speak
nonliterally. Its principal function, however, is that it allows speakers
to make use of contextual features to speak far more concisely than
otherwise. Imagine how lengthy utterances would be if everything we
wanted to express had to be spelled out explicitly in the grammar of
our sentences. Pragmatics saves us from this wasteful verbosity. Thus
instead of using sentences like (1), we can, on occasion, use sentences
like (2)
(1)
The man who just asked the stupid question about the relation
betwen the mental and the physical has, thank God, left the room
(2)
Thank God, he's gone
(Katz 1977: 19-20)
VI. Expressibility and indexicality
It is ironic that Katz used the 'Thank God' example, for many years before
(in 1959) Arthur Prior had published an article entitled 'Thank Goodness
that's over', in order to support the opposite conclusion: that there are
sentences whose content cannot, even 'in principle', be made fully explicit
in a context-independent manner:
One says, e.g. 'Thank goodness that's over', and not only is this, when
said, quite clear without any date appended, but it says something
which it is impossible that any use of a tenseless copula with a date
should convey. (Prior 1959: 84; emphasis mine)
In contrast to Prior, Katz insists that reliance on contextual clues can be
dispensed with, in principle if not in fact. That follows from the Principle
of Effability, and it seems to follow from the Principle of Expressibility as
well. Yet Searle, contrary to Katz, does not say so explicitly. He expresses
no firm views on these matters but seems to oscillate between two
positions.
On the one hand Searle allows that one way of making explicit who
one means by e.g. the pronoun 'he' is to demonstrate the referent — to
point to him. The ability to provide a demonstrative identification of the
referent in context counts as satisfying the Principle of Expressibility, he
says, just as much as the ability to provide a description such as 'The man
who just asked the stupid question about the relation betwen the mental
and the physical'.
Applied to the present case of definite reference, [the Principle of
Expressibility] amounts to saying that whenever it is true that a
speaker means a particular object... it must also be true that he can say
exactly which object it is that he means. (...) A limiting case of saying
is saying which involves showing; that is, a limiting case of
satisfying... the principle of expressibility is indexical presentation of
the object referred to. (Searle 1969: 88)
On the other hand, Searle also speaks as if the contextual demonstration
itself was a way of 'communicating', without making fully explicit, a sense
that could be made fully explicit by replacing the demonstration by
linguistic symbols. Since the pointing gesture is not part of the linguistic
'expression' but part of the 'context', the ability to articulate the sense of the
pointing gesture in words is part and parcel of what the Principle of
Expressibility requires. Thus, a sentence such as 'The man (or: that man) is
a foreigner' (accompanied by a glance or a pointing gesture) could be
rephrased more explicitly as 'There is one and only one man on the
speaker's left by the window in the field of vision of the speaker and the
hearer, and he is a foreigner' (Searle 1969: 92).
Be that as it may, the Principle of Expressibility can and should be
weakened so as not to entail that indexicality is eliminable. One way of
doing so is to broaden the notion of 'sentence meaning' so as to admit
among determinants of sentence meaning contextual assignments of
semantic values to indexical expressions. Thus interpreted the principle of
Expressibility is no longer equivalent to the Principle of Effability; it does
not say that the content of the speech act can always be fully encoded into
the linguistic meaning of a sentence type, but that it can be literally
expressed by a sentence type 'with respect to context', where context
consists of a time of utterance, a place of utterance, a speaker, a hearer, a
sequence of demonstrated objects, etc. (That is the 'context' in the narrow
sense in which formal semanticists use the notion.)
Thus weakened, the Principle of Expressibility is still incompatible
with Searle's later findings about the background. What those findings
show is that, however explicit the sentence, its linguistic meaning does not
determine a set of truth conditions even 'with respect to context'. To
account for the phenomena adduced by Searle in his later writings, we
would have not only to relativize sentence-meaning to context but also to
broaden the notion of context so as to include the total 'background'. If we
broaden the notion of context that way, however, we fall back on the
interpretation of the Principle of which I said that it is obviously not what
Searle intended. An utterance is not explicit, by Searle's early standards, if
it is an utterance of a vague or ambiguous sentence. But if the 'context', in
the richest possible sense, is allowed to compensate for the lack of
determinacy of the sentence, then even the utterance of a vague or
ambiguous sentence will count as explicit. That is clearly not what Searle
originally meant when he talked of an utterance being explicit (or not).
VII. Local expressibility, global inexpressibility
Let us go back to the passage in which Searle says that the Principle of
Expressibility is compatible with background phenomena:
There is nothing in the thesis of the relativity of literal meaning which
is inconsistent with the Principle of Expressibility, the principle that
whatever can be meant can be said. It is not part of, nor a
consequence of, my argument for the relativity of literal meaning that
there are meanings that are inherently inexpressible. (Searle 1979:
134)
The last sentence suggests another possible weakening of the Principle of
Expressibility. The principle could be understood as saying simply that
whatever is meant can be made explicit. That entails that every
background assumption relied upon in interpreting an utterance can be
made explicit, but this is compatible with the fact that (i) they cannot be all
be made explicit at the same time, and (ii) whenever we make one
assumption explicit by adding more descriptive material, further
background assumptions are implicitly called upon for the interpretation of
that extra material. The Principle of Expressibility thus weakened becomes
a Principle of Local Expressibility. In one passage in Speech Acts Searle
seems to have had such a weak version in mind:
Another application of this law [the Principle of Expressibility] is that
whatever can be implied can be said, though if my account of
preparatory conditions is correct, it cannot be said without implying
other things. (Searle 1969: 68-69; emphasis mine)
Even that weakening is not satisfactory, however. The Principle of
Expressibility, thus weakened, no longer supports the claim that "the study
of sentence meanings and the study of speech acts are one and the same
study". If expressibility can only be local, then a principle of global
inexpressibility also holds, according to which what is said explicitly is
always said against a background of unarticulated assumptions. That is
sufficient to justify the contextualist claim that there is more to the content
of a speech act than can be encoded into the meaning of a sentence. But
Searle used the Principle of Expressibility precisely to argue against such a
view.
Before concluding that Searle was mistaken when he said that the
phenomenon of background-dependence does not refute the Principle of
Expressibility, there is a last option that should be tried. I think it may well
be what Searle had in mind.
VIII. The generalization of background-dependence to all Intentional
states
Searle says that what is true of linguistic meaning is true of all Intentional
states: thoughts, perceptions, intentions, etc. In all cases the Intentional
content of the state determines satisfaction-conditions only relative to
background assumptions which cannot be realized as further aspects of
that content. If this is right, then there is a sense in which there may well
be a perfect fit between the meaning of the sentence (which determines
conditions of satisfaction only against a background of assumptions) and
what the speaker means by uttering the sentence (since the speaker's
meaning intentions themselves are background-relative in just the same
way). In other words a sentence can be explicit, in the sense that it
corresponds exactly to what the speaker means, without ceasing to underdetermine the conditions of satisfaction of the speech act. On that view the
content of the speech act is the content of the sentence; both underdetermine the conditions of satisfaction. The Principle of Expressibility is
therefore satisfied despite the phenomenon of background-dependence.
Absolute explicitness is impossible since background-dependence is
ineliminable, but relative explicitness can be achieved, consistently with
the Principle of Expressibility. By 'relative' explicitness I mean a perfect fit
between (i) the semantic content of the sentence, (ii) the content of the
speech act performed by uttering the sentence, and (iii) the content of the
Intentional states expressed by the utterance.4 Thus when I say 'The cat is
on the mat', the literal meaning of the sentence (with respect to a
contextual assignment of values to the incomplete descriptions 'the cat' and
'the mat') is the same thing as the content of the assertion that the cat is on
the mat, and that is identical to the content of the expressed belief that the
cat is on the mat. Background-dependence applies in all three cases; and it
applies to the perception that the cat is on the mat as well:
In my present experience I assume that I am perceiving the cat and
the mat from a certain point of view where my body is located; I
assume that these visual experiences are causally dependent on the
state of affairs that I perceive; I assume that I am not standing on my
head and seing cat and mat upside down, etc.; and all these
assumptions are in addition to such general assumptions as that I am
in a gravitional field, there are no wires attaching to cat and mat, etc.
Now, the Intentionality of the visual experience will determine a set
of conditions of satisfaction. But the purely visual aspects of the
experience will produce a set of conditions of satisfaction only
against a set of background assumptions which are not themselves
part of the visual experience... In this case as in the literal meaning
case, the Intentionality of the visual perception only has an
application, only determines a set of conditions of satisfaction,
against some system of background assumptions. (Searle 1979: 136)
To sum up: what one literally says depends upon the Background,
but what one believes and what one perceives also depend upon the
Background. In all cases the content of the representation — be it
linguistic or mental — only determines conditions of satisfaction against a
background of unarticulated assumptions. The question, whether the
beliefs one communicates can be exactly expressed by the sentences one
As far as (ii) and (iii) are concerned, Searle points out that, in virtue of the theory of
speech acts, the content of the speech act always corresponds to the content of the
4
Intentional state it expresses.
utters, can therefore be answered affirmatively, in accord with the
Principle of Expressibility, even though the uttered sentence can't be fully
explicit in the absolute sense.
That view, which it is reasonable to ascribe to Searle, stands in sharp
contrast to an alternative position, deriving from Wittgenstein. The
alternative position sets linguistic meaning apart from Intentional states: It
says that words are special in being inert and (as Searle himself insists)
devoid of 'intrinsic Intentionality'. What gives them 'life' is the use that is
made of them. There is no such thing for Intentional contents. In contrast
to words and sentences, thoughts and concepts are not 'tools', and they are
not 'used'. Accordingly they lack the semantic indeterminacy which
characterizes sentences and linguistic material generally. While sentences
are semantically indeterminate except in the context of a speech act, 5
thoughts are semantically determinate. Hence it is a category mistake to
generalize, as Searle does, the sort of contextual dependency exhibited by
linguistic meaning to Intentional states in general.
In the remainder of this paper, I will, first, scrutinize the view I have
(tentatively) ascribed to Searle: that which generalizes the phenomenon of
background-dependence to all Intentional states and is thereby able to
protect the Principle of Expressibility. I will show that the attempted
generalization fails. I will then elaborate the Wittgensteinian position and
show that it can accommodate the phenomena adduced by Searle in his
critique of the Determination View.
IX. Literal meaning vs. Intentional content
Searle's view rests on the following equation:
According to James Conant, this Wittgensteinian principle is a generalization of
Frege's celebrated Context principle: "[Wittgenstein] seeks to genralize Frege's contextprinciple so that it applies not only to words (and their role within the context of a
significant proposition) but to sentences (and their role within the context of
5
circumstances of significant use)" (Conant 1998: 233).
sentence
meaning
————
conditions of
satisfaction
of utterance
=
content of
speech act
————
conditions of
satisfaction
of speech act
=
content of
Intentional
state
————
conditions of
satisfaction
of Int. state
Table 2
If that equation could be maintained, it would indeed be possible to
conciliate the Principle of Expressibility and the phenomenon of
background-dependence. But I do not think the equation can be
maintained, for the following reason.
Even if 'sentence meaning' is understood as the meaning of the
sentence with respect to contextual assignments of values to indexicals, it
is still much more indeterminate, much more susceptible to background
phenomena, than the content of the speech act or the content of the
expressed psychological state. There is this basic difference between the
two sorts of case: If we change the background while keeping the meaning
of the sentence constant, we change the truth conditions — that is what
Searle's examples show; but we simply cannot, by manipulating the
background, change the conditions of satisfaction of the speech act or of
the Intentional state while leaving its content unchanged. The content of
the speech act (or of the Intentional state) lacks the form of 'indeterminacy'
which the meaning of the sentence possesses, and which makes it possible
to keep it constant while varying the conditions of satisfaction.
Searle's formulations are misleading in that respect. For he
repeatedly says that the content of a speech act, or the content of an
Intentional state, only determines conditions of satisfaction against a
background of unarticulated assumptions, just as the meaning of the
sentence only determines conditions of satisfaction against the
Background. But in the case of speech acts and Intentional states, the
relevant 'contents' are not separable from the conditions of satisfaction
they determine. The order to cut the grass is not the same order when 'cut'
is understood as 'slice' and when it is understood as 'mow'. That is so
because you can't change the conditions of satisfaction (by manipulating
the background) without eo ipso changing the content and therefore (since
the act/state is individuated in part by its content) without changing the
state or the act itself.
The inseparability of content from conditions of satisfaction shows
up everywhere in Searle's writings. Here are a few quotations from
Intentionality:
An Intentional state only determines its conditions of satisfaction —
and thus only is the state that it is — given its position in a Network
of other Intentional states and against a Background of practices and
preintentional assumptions that are neither themselves Intentional
states nor are they parts of the conditions of satisfaction of Intentional
states. (Searle 1983: 19)
The Intentional content which determines the conditions of
satisfaction is internal to the Intentional state: there is no way the
agent can have a belief or a desire without it having its conditions of
satisfaction. (Searle 1983: 22)
Intentional contents in general and experiences in particular are
internally related in a holistic way to other Intentional contents (the
Network) and to nonrepresentational capacities (the Background).
They are internally related in the sense that they could not have the
conditions of satisfaction that they do except in relation to the rest of
the Network and the Background. (Searle 1983: 66)
Intentional states only have the conditions of satisfaction that they do,
and thus only are the states that they are, against a Background of
abilities that are not themselves Intentional states. (Searle 1983: 143)
The following passage is particularly interesting:
It would... be incorrect to think of the Background as forming a
bridge between Intentional content and the determination of
conditions of satisfaction, as if the Intentional content itself could not
reach up to the conditions of satisfaction. (Searle 1983: 158)
What is interesting here is the contrast with literal meaning. For in the
case of literal meaning, there is a clear sense in which the meaning of the
sentence itself 'does not reach up to the conditions of satisfaction'. Searle
says so in many places. For example: "If somebody instructs me to cut the
sand, I do not know what I am supposed to do. For each [such] case ['cut
the sand', 'cut the mountain', etc.] I can imagine a context in which I would
be able to determine a set of truth conditions; but by themselves, without
the addition of a context, the sentences do not do that." (Searle 1980: 2256.) The sentence, with its meaning, can easily be separated from the
conditions of satisfaction which, in context, it determines. No so with
Intentional states (or speech acts) and their contents.
In general, to convince ourselves that the above equation can't be
maintained, there is a very simple procedure: one has only to consider
what happens if we replace 'sentence' by 'Intentional state' (or 'speech act')
and 'literal meaning' by 'content' in one of the numerous passages in which
Searle describes the under-determination of truth conditions by literal
meaning. The results are instructing. Here is one example:
Original passage
The literal meaning of a sentence or expression only determines a set
of truth conditions given a set of background assumptions and
practices. Given one set of these a sentence or expression may
determine one set of truth conditions and given another set of
assumptions and practices the same sentence or expression with the
same meaning can determine a different set of truth conditions.
(Searle 1980: 227)
Same passage after substitution:
The content of a speech act or Intentional state only determines a set
of truth conditions given a set of background assumptions and
practices. Given one set of these a speech act or Intentional state may
determine one set of truth conditions and given another set of
assumptions and practices the same speech act or Intentional state
with the same content can determine a different set of truth
conditions.
In view of the inseparability thesis, the claim that 'the same speech act or
Intentional state with the same content can determine different sets of truth
conditions' is nonsense. Again, if you change the conditions of
satisfaction, the content does not stay constant. So the content of the
speech act or Intentional state does not play the same role, and does not
have the same properties, as the meaning of the sentence; for it is crucial to
Searle's argument in 'Literal meaning' and elsewhere that the meaning of
the sentence stays constant when the truth conditions are made to vary by
manipulating the background.
X. Literal meaning, sensory content, and the brain
When he stresses the analogy between literal meaning and Intentional
content, Searle often appeals to the example of perception. The following
passage is characteristic:
All of the arguments for the context dependency of the sentences
"Bill cut the grass" , "4 + 5 = 9" and "Snow is white" are also
arguments for the context dependency of the beliefs that Bill cut the
grass, that 4 + 5 = 9 and that snow is white. The content of those
beliefs determines the conditions of satisfaction that they do
determine only against a background. "Well", we might imagine our
objector saying, "if so that is because those beliefs would naturally
come to us in words. But how about worldless Intentional states, and
how about the primary form of Intentionality, perception?" If
anything the contextual dependency of perceptual contents is even or
more striking [sic] than the contextual dependency of semantic
contents. Suppose I am standing in front of a house looking at it; in so
doing I will have a certain visual experience with a certain Intentional
content, i.e. certain conditions of satisfaction; but suppose now as
part of the background assumptions I assume I am on a Hollywood
movie set and all of the buildings are just papier maché façades. This
assumption would not only give us different conditions of
satisfaction; it would even alter the way the façade of the house looks
to us, in the same way that the sentence "Cut the grass!" would be
interpreted differently if we thought that the background was such
that we were supposed to slice the grass rather than mow it. (Searle
1980: 231)
Perception indeed supports the analogy to some extent. Even in that case,
however, Searle acknowledges that the content of the visual experience
changes when the background is altered:
It is part of the content of my visual experience when I look at a
whole house that I expect the rest of the house to be there if, for
example, I enter the house or go around to the back. In these sorts of
cases the character of the visual experience and its conditions of
satisfaction will be affected by the content of the beliefs that one has
about the perceptual situation. I am not going beyond the content of
my visual experience when I say, "I see a house" instead of "I see the
façade of a house", for, though the optical stimuli may be the same,
the conditions of satisfaction in the former case are that there should
be a whole house there. (Searle 1983: 54-55)
If the content of the visual experience changes when the conditions
of satisfaction are manipulated by altering the background, is there
something that stays constant and can be compared to the constant
meaning of the sentence? Searle names two candidates: 'the purely visual
aspects of the experience' and 'the optical stimuli'. Now the optical stimuli
are not a good candidate. To use a contrast made famous by John
McDowell (1984: 103n), they may be a bearer of content, but they are not
an aspect of content. What we need, for the analogy with literal meaning
to hold, is an aspect of semantic content that stays invariant when
background assumptions are manipulated. The 'purely visual aspects of the
experience' seem to fit the bill. It is common to distinguish two forms of,
or two levels in, perception. Cognitive perception is higher-level
perception, and it presupposes a lower level of sensory perception.
Sensory perception is modular and unaffected by background knowledge;
cognitive perception, is nonmodular and background-dependent (see e.g.
Dretske 1990: 138-146). If the distinction is sound, the content of sensory
perception corresponds to what Searle calls the 'purely visual aspects of
the experience', and that is indeed analogous to the linguistic meaning of
the sentence.
The problem is that the distinction between the aspects of visual
content which are modular and those which are cognitive and backgrounddependent cannot be generalized to all Intentional states. There is no such
contrast for beliefs, desires or intentions. Nor is there such a contrast for
thought in general. The distinction seems to be limited to perceptual states
and processes. It is indeed similar to the distinction we find in the
language case, but that similarity itself does not provide an explanation of
the phenomenon of background-dependence in the language case; it rather
constitutes a further fact in need of explanation (a fact which I will leave
aside in this chapter).
Searle mentions a third candidate for the analogy with literal
meaning; a candidate that has the relevant degree of generality. The neural
configuration in the brain which realizes a given Intentional state can stay
constant even though we radically alter the Background. For example, take
Carter's desire to run for the Presidency of the US and the corresponding
neural configuration in Carter's brain. We can suppose that "exactly these
same type-identical realizations of the mental state occurred in the mind
and brain of a Pleistocene man living in a hunter-gatherer society of
thousands of years ago" (Searle 1983: 20). Because of the dependence of
Intentional contents on Network and Background, "however type-identical
the two realizations might be, the Pleistocene man's mental state could not
have been the desire to run for the Presidency of the United States" (id.).
Granted. It is well known that content, in general, is not an intrinsic but a
relational property of the content-bearing state. That is the lesson of
Externalism. Had the environment been sufficiently different, the same
neural state which realizes a given Intentional content would realize a
different content (or no content at all). But this is not the same
phenomenon as background-dependence. The neural state which realizes a
given Intentional content is not an aspect or level of content; it is, again, a
vehicle, a bearer of content. As such it is analogous to the sentence qua
syntactic object, rather than to the linguistic meaning of the sentence.
What corresponds to Externalism in the linguistic case is therefore the fact
that the sentence (type) could mean something different from what it
actually means: it would do so if the conventions of the language had been
different. This has nothing to do with the under-determination of semantic
content (given a fixed linguistic meaning). Similarly, the fact that a neural
state realizes a given Intentional content only in a certain context does not
show that in thought, as in language, there is a level of content that underdetermines conditions of satisfaction.
I conclude that the analogy between the background-dependence of
Intentional content and the background-dependence of semantic content
breaks down at crucial points and does not, as it stands, provide an
explanation for the facts adduced by Searle in his critique of the
Determination View. I therefore suggest that, getting rid of the Principle of
Expressibility, we turn to the contextualist approach and see what can be
done within that framework.
XI. A contextualist perspective
The account of the phenomenon of background-dependence I am about to
provide takes its inspiration from Austin's theory of truth (cf. the paper
'Truth' in Austin 1971) and, above all, from the remarks of Waismann
(1951) on the open texture of empirical predicates; which remarks
themselves presumably echo Wittgenstein's views (see, in particular,
sections §§66sq. of Philosophical Investigations). The central idea is that
words are not primitively associated with abstract 'conditions of
application', constituting their conventional meaning (as on the Fregean
picture). Rather, they are associated with particular applications.
Consider what it is to learn a predicate P. The learner, who I'll call
Tom, observes the application of P in a particular situation S; he associates
P and S. At this stage, the 'meaning' — or, as I prefer to say, semantic
potential — of P for Tom is the fact that P is applicable to S. In a new
situation S', Tom will judge that P applies only if he finds that S'
sufficiently resembles S. To be sure, it is possible that S' resembles S in a
way which is not pertinent for the application of P. The application of P to
S' will then be judged faulty by the community, who will correct Tom. The
learning phase for Tom consists in noting a sufficient number of situations
which, like S, legitimate the application of P, as opposed to those, like S',
which do not legitimate it. The semantic potential of P for Tom at the end
of his learning phase can thus be thought of as a collection of legitimate
situations of application; that is, a collection of situations such that the
members of the community agree that P applies to those situations. Let's
call the situations in question source-situations. The future applications of
P will be underpinned, in Tom's usage, by the judgement that the situation
of application (or target-situation) is similar to the source-situations.
In this theory the semantic potential of P is a collection of sourcesituations, and the conditions of application of P in a given use, involving
a given target-situation S'', are a set of features which S'' must possess to
be similar to the source-situations. The set of features in question, and so
the conditions of application for P, will not be the same for all uses; it is
going to depend, amongst other things, on the target-situation. One targetsituation can be similar to the source-situations in certain respects and
another target-situation can be similar to them in different respects. But the
contextual variability of the conditions of application does not end there.
Even once the target-situation is fixed, the relevant dimensions for
evaluating the similarity between that situation and the source-situations
remain under-determined: those dimensions will vary as a function of the
subject of conversation, the concerns of the speech participants, etc.
One particularly important factor in the contextual variation is the
relevant 'contrast set'. As Tversky (1977) has pointed out, judgements of
similarity are very much affected by variations along that dimension. If we
ask which country, Sweden or Hungary, most resembles Austria (without
specifying the relevant dimension of similarity), the answer will depend on
the set of countries considered. If that set includes not just Sweden,
Hungary and Austria but also Poland, then Sweden will be judged more
like Austria than Hungary; but if the last of the four countries considered
is Norway and not Poland, then it is Hungary which will be judged more
like Austria than Sweden. The explanation for that fact is simple. Poland
and Hungary have certain salient geopolitical features in common which
can serve as basis for the classification: Hungary and Poland are then put
together and opposed to Austria and Sweden. If we replace Poland by
Norway in the contrast set a new principle of classification emerges, based
on the salient features shared by Norway and Sweden: in this new
classification Hungary and Austria are back together. Tversky concludes
that judgements of similarity appeal to features having a high 'diagnostic
value' (or classificatory significance), and that the diagnostic value of
features itself depends on the available contrast set.
XII. Accounting for background-dependence
Within that simple contextualist framework, let us reconsider the
phenomenon of background-dependence. It goes along with the global
character of the similarity between target-situation and source-situations.
The source-situations are concrete situations with an indefinite number of
features. Some of these features are ubiquitous and their diagnostic value
in a normal situation is vanishing.6 They belong to the most general and
immutable aspects of our experience of the world: gravity, the fact that
food is ingested via the mouth, etc. When we specify the truth conditions
of a sentence (for example the sentence 'The cat is on the mat'), or the
conditions of application of a predicate (for example the predicate 'on' in
that sentence), we only mention a small number of features — the
'foreground' features — because we take most of the others for granted; so
Cf. Tversky 1977: 342: "The feature 'real' has no diagnostic value in the set of actual
animals since it is shared by all actual animals and hence cannot be used to classify
them. This feature, however, aquires considerable diagnostic value if the object set is
6
extended to include legendary animals, such as a centaur, a mermaid or a phoenix."
we do not mention gravity, we presuppose it. Nevertheless, gravity is one
of the features possessed by the situations which are at the source of the
predicate 'on'; and there is an indefinite number of such features. These
background features of the source-situations can be ignored inasmuch as
they are shared by the situations of which we can want to speak when we
utter the sentence; but if we imagine a target-situation where the normal
conditions of experience are suspended, and where certain background
features of the source-situations are not present, then we shatter the global
similarity between the target-situation and the source-situations. Even if
the target-situation has all the foreground features which seem to enter
into the 'definition' of a predicate P, it suffices to suspend a certain
number of background features in order to jeopard the application of P to
the target-situation. That shows that the semantic potential of P is not, as
in Fregean semantics, a set of conditions of application determined once
and for all, but a collection of source-situations such that P applies to a
target-situation if and only if it is relevantly similar to the sourcesituations.
A caveat: as Searle himself emphasizes, the fact that the targetsituation does not possess certain background features of the sourcesituations does not automatically entail the non-applicability of the
predicate P. It can be that the background features which the targetsituation does not possess (for example gravity) are contextually irrelevant
and do not affect the application conditions of the predicate. For the same
sort of reason, the possession by the target-situation of what I have called
the foreground features of the source-situations is no more a necessary
condition for the application of the predicate than it is a sufficient
condition. For a predicate (or a sentence) to apply to a target-situation that
situation must resemble the source-situations under the contextually
relevant aspects. So a predicate can apply even if the target-situation
differs markedly from the source-situations, as long as, in the context and
taking into account the contrast set, the similarities are more significant
than the differences. Thus, in certain contexts, the predicate 'lemon' will
apply to plastic lemons, or the word 'water' to XYZ. Putnam himself, in
'The Meaning of Meaning', recognizes the legitimacy of such uses, made
possible by the contextual variability of the relevant dimensions of
similarity (Putnam 1975: 238-9).
XIII. Conclusion
I take the phenomenon of background-dependence to reveal quite
fundamental features of natural language. Searle must be credited for
having drawn attention to that phenomenon and for having appreciated its
importance. I have criticized Searle's explanation of the phenomenon,
however. According to Searle, the under-determination of semantic
content is a special case of a more general phenomenon which affects all
representations, whether linguistic or mental. To determine whether or not
a representation is 'satisfied', that representation must be interpreted.
Searle cites the Wittgensteinian example of an image showing a man
climbing a slope: the man could just as well be seen as going backwards
down the slope — the image itself does not tell us which interpretation is
right (Searle 1992: 177). For Searle, the under-determination of
satisfaction-conditions derives from the fact that representations, whether
linguistic or mental, are not 'self-interpretive' or 'self-applicative'. From
that follows the non-representational character of the Background which
bridges the gap between the representation and its application. To add a
second representation to the first in order to interpret it does no more than
postpone the problem, for the second representation would also need
interpreting. Ultimately, a representation can only be applied if it is
inserted in a nonrepresentational milieu — if it plays a role in a practice.
Whence Searle's insistence on the fact that the Background consists largely
in behavioural dispositions and know-how. "Intentionality occurs in a
coordinated flow of action and perception, and the Background is the
condition of possibility of the forms taken by the flow" (Searle 1992: 195).
What we assume we assume in virtue of the way we act and navigate
through the world. We assume gravity, the solidity of objects, and the
existence of other minds, in virtue of the way we act; we do not, or need
not, entertain thoughts about these things.
I take Searle to be right both concerning the need for interpretation
and the importance of the 'practical' dimension of cognition. But I doubt
the two things are related in the way Searle makes them appear to be.
Moreover, I do not think we can simply invoke the non-self-interpretive
(or non-self-applicative) character of representations; we must explain it.
Why do the representations conveyed by words only apply to the world via
a process of interpretation? Why aren't they self-applicative? They ought
to be, if linguistic meaning conformed to the Fregean image, that is, if it
consisted in conditions of application. If, in virtue of the conventions of
the language, a predicate P possesses definite conditions of application, as
the Fregean thinks it does, then either the reality of which we speak
satisfies those conditions and the predicate applies, or it does not satisfy
them and the predicate does not apply. I grant the non-self-applicative
character of linguistic representations as an empirical datum, attested by
Searle's examples, but to give an account of that feature an alternative
must be proposed to the traditional view of meaning inherited from Frege.
In section XI, I sketched such an alternative.
Searle's generalization of background-dependence to all Intentional
states enables him to save the Principle of Expressibility which is the heart
of his earlier theory of speech acts. That move I do not find very
convincing, for I have always been struck by the tension between the
earlier philosophy, based on the Principle of Expressibility, and the later
views which pull in the opposite direction. Be that as it may, I have shown
that the attempted generalization fails. It follows that the Principle of
Expressibility cannot be saved. More important, we are left without an
explanation of the phenomenon of background-dependence. Where does it
come from?
Impressed by the similarity between Searle's backgrounddependence and Waismann's 'open texture', I have offered a contextualist
account of background-dependence. On this view the content or sense of
words — their contributions to the truth conditions of utterances — must
be contextually constructed in an active process of interpretation; it is not
ready-made. What is given as part of the language is not the sense of
words, which must be constructed, but only what I have called their
semantic potential. To construct the (context-dependent) sense of a word
out its (context-independent) semantic potential, nothing short of the full
situation of utterance will do. An impoverished 'context' consisting only of
values for a given set of parameters does not provide the sort of input
which is needed for the process of sense construction to get off the ground;
for that process relies on a global assessment of similarity between
situations possessing, in principle, an indefinite number of features.
References
Austin, J.L. (1971), Philosophical Papers, 2nd ed. Oxford: Clarendon
Press.
Conant, J. (1998), Wittgenstein on Meaning and Use. Philosophical
Investigations 21: 232-250.
Dretske, F. (1990), Seeing, Believing, and Knowing, in D. Osherson et al.
(eds.), An Invitation to Cognitive Science, vol. 3, pp. 129-148. Cambridge,
Mass: MIT Press.
Forguson, L. (1973), Locutionary and Illocutionary Acts, in G. Warnock
(ed.), Essays on J. L. Austin, pp. 160-185. Oxford: Clarendon Press.
Geach, P. (1972), Logic Matters. Oxford: Basil Blackwell.
Grice, P. (1989), Studies in the Way of Words. Boston: Harvard University
Press.
Katz, J.J. (1972), Semantic Theory. New York: Harper & Row.
Katz, J.J. (1977), Propositional Structure and Illocutionary Force. New
York: Crowell.
Katz, J.J. (1978), Effability and Translation, in F. Guenthner and M.
Guenthner-Reutter (eds.), Meaning and Translation, pp. 191-234. New
York: New York University Press.
McDowell, J. (1984), De Re Senses, in C. Wright (ed.), Frege: Tradition
and Influence, pp. 98-109. Oxford: Basil Blackwell.
Prior, A. (1959), Thanks Goodness That's Over, reprinted in his Papers in
Logic and Ethics, pp. 78-84. London: Duckworth, 1976.
Putnam, H. (1975), The Meaning of Meaning, in his Mind, Language and
Reality: Philosophical Papers, vol. 2, pp. 215-271. Cambridge: Cambridge
University Press.
Recanati, F. (1980), Qu'est-ce qu'un acte locutionnaire? Communications
32: 190-215.
Recanati, F. (1987), Meaning and Force. Cambridge: Cambridge
University Press.
Recanati, F. (1994), Contextualism and anti-contextualism in the
philosophy of language, in S. Tsohatzidis (ed.), Foundations of Speech Act
Theory, pp. 156-166. London: Routledge.
Searle, J. (1965), What is a Speech Act?, in M. Black (ed.), Philosophy in
America, pp. 221-239. Ithaca: Cornell University Press.
Searle, J. (1968), Austin on Locutionary and Illocutionary Acts, reprinted
in G. Warnock (ed.), Essays on J. L. Austin, pp. 141-159. Oxford:
Clarendon Press, 1973.
Searle, J. (1969), Speech Acts. Cambridge: Cambridge University Press.
Searle, J. (1971), Introduction, in J. Searle (ed.), The Philosophy of
Language, pp. 1-12. Oxford: Oxford University Press.
Searle, J. (1975), Indirect Speech Acts. Syntax and Semantics 3: 59-82.
Searle, J. (1978), Literal Meaning. Erkenntnis, 13: 207-24, reprinted in
Searle (1979) pp. 117-136.
Searle, J. (1979), Expression and Meaning. Cambridge: Cambridge
University Press.
Searle, J. (1980), The Background of Meaning, in J. Searle, F. Kiefer and
M. Bierwisch (eds.), Speech Act Theory and Pragmatics, pp. 221-32.
Dordrecht: Reidel.
Searle, J. (1983), Intentionality. Cambridge: Cambridge University Press.
Searle, J. (1992), The Rediscovery of the Mind. Cambridge, Mass: MIT
Press.
Tversky, A. (1977), Features of Similarity. Psychological Review 84: 327352.
Waismann, F. (1951), Verifiability, in A. Flew (ed.), Logic and Language,
1st series, pp. 117-144. Oxford: Basil Blackwell.
Wittgenstein, L. (1953), Philosophical Investigations, Oxford: Basil
Blackwell.